The objectives of this study were to determine the cumulative incidence and timing of new brain metastases over the course of systemic therapy for metastatic melanoma and to identify prognostic factors for brain metastases. Chemo-naive patients underwent computed tomography or MRI of the brain every 6 weeks. The cumulative incidence of confirmed brain metastases was calculated at 12-week intervals. Univariable and multivariable competing risk regression models were used to assess the association between the development of brain metastases and potential risk factors of interest. Cumulative incidence with competing risk and competing risk regression was used to assess the brain metastasis-free interval from the time of diagnosis of stage IV disease. The clinical characteristics of the 315 patients with brain metastases were compared with those of 370 brain metastasis-free patients. Among patients with brain metastases, a significantly higher proportion had stage M1b and M1c disease at diagnosis compared with stage M1a and a greater proportion had metastatic disease in three or more visceral sites. Significantly shorter brain metastasis-free intervals were found in these patients compared with patients with M1a disease and those with no visceral metastases. More than 80% of the 230 patients who developed brain metastases during systemic therapy had their brain metastases confirmed within 60 weeks from the onset of advanced melanoma. Imaging studies at 12-week intervals for 60 weeks after the diagnosis of advanced melanoma will detect brain metastases in most of the patients who will eventually develop them.